Test a hypothesis of a polynomial model order in design of an experiment tasks

A criterion of test a hypothesis of a polynomial regression order is substantiated, in it the resultant statistics is formed by using two designs. The first supposes an independent variable to take a set of equidistant points and the second design is optimal. As a criterion of the design optimality the minimum of a sum of quadrates deviations of the regression function estimate from it truly values on a given point set is used. The examples of using the developed criterion in the task of finding performances of telecommunication systems are given.